How to find AI ROI in IT and customer support

Recent studies from Laks Srinivasan’s advisory firm—the ROI of AI Institute—show that 90% of organizations are struggling to generate business value from their investments in AI. One key differentiator he sees from the successful 10% is what he calls “strategic intent.”

That is, companies that align AI pilot projects with clear business objectives are far more likely to see tangible results.

And the same rule applies within the enterprise, says Srinivasan. Leaders in two internal hotbeds of AI adoption—IT organizations and customer support teams—can up their odds of healthy ROI, as well as reduce risk and cost, by putting strategic intent into practice.  

In a recent interview, Srinivasan, the RoAI Institute’s co-founder and managing director, explained some of the biggest ROI opportunities with AI he sees in IT and customer support. 

What are some basic principles for maximizing returns on AI investments, especially in two huge areas of need: IT and customer support? 

It’s important to start with outcomes first and then AI next. If you start with AI use cases first, that’s a solution chasing a problem. The batting averages on innovation go up substantially when you start with outcomes first.

In our research, we find that somewhere between 5% and 10% of companies say that they’re getting meaningful value from AI. What are they doing differently? They have a very disciplined, systematic way to identify high-value, high-impact use cases and prioritize them.

They know their customers. They know what the pain points are and where the opportunities are. For IT, it’s the same methodology because employees are their internal customers. They’re able to be very precise in behavior-based customer segmentation and identifying specific outcomes where AI could offer a solution.

What we advise is that AI is not a broad-brush technology. These are not moonshots. You want to look at what AI allows you to do. The companies that are getting real value from AI are finding smaller and smaller kinds of problems that they can deploy AI against. Once they identify those outcomes, companies should look at ROI through both a tactical and strategic lens.

In customer service and support, how do you define tactical vs. strategic returns?

From our vantage point, tactical returns might be call-center volume reduction, eliminating the need for the customer to call in on certain types of calls, and once the customer calls, reducing the amount of time it takes to resolve an issue.

For example, a telecom media company we worked with found that call volumes spiked around certain days of the month, particularly when customers received their bills. Applying AI, they were able to predict, on an individual customer basis, when millions of customers might call and what questions that they’re calling about. They could then automatically send out an email or a message over another digital channel to exactly answer that question.

With tactical applications of AI, you can see quick returns. In that case, we did a 90-day market test and then rolled it out and saw a 50% reduction in calls about bills over six to nine months.

Strategic returns are more around improving customer satisfaction because that improves customer lifetime value. Loyalty has a lot of knock-on effects and improves your competitive position, but it might take longer to measure that value.

What are the biggest ROI opportunities in customer support?

What I see these days is when business leaders talk about AI, they mean generative AI. But for the average enterprise, the majority of the value they’re going to get from AI is going to be in traditional or classic AI. For the telecom media company we were discussing, it was predictive analytics that enabled them to reduce call volume by 50%, not generative AI.

That’s not to say generative AI isn’t important. Over 90% of companies are experimenting with it, and they should be experimenting. We see real potential value around customer service chatbots if you could deploy them correctly. It won’t completely eliminate a contact center, but it could potentially field certain types of calls 24/7, expanding the time window in which your customers can interact with the brand and the company.

But we are not yet finding a lot of companies are able to go beyond initial proof points because they haven’t defined a desired outcome opportunity around a particular KPI to realize it at scale. It’s still early days.

How about IT organizations—what are the high-value AI opportunities they should be pursuing?

Based on companies we’ve worked with, there are a lot of tactical opportunities for AI to improve efficiencies with helpdesk ticket volumes by identifying and predicting which employee is the most likely to call and seek help, and why.

I’ve seen use cases where you can personalize and customize an employee’s onboarding journey. A good chunk of level-one calls are around password resets. Strategically, for IT, you have the same opportunities in improving employee satisfaction and loyalty as you do with external customers, but those outcomes don’t tend to receive the same level of management focus as call-center AI cases.

Recently, I had a room full of executives of a Fortune 500 company, including the CIO, as part of a workshop, and everybody said, “We wish we could improve employee helpdesk, but employee-facing initiatives are not as big of a priority as customer support.” There’s value to be had from AI there, but they’re hard-pressed to make AI investment a priority, given that IT is a cost center. And while in bigger companies, CIOs clearly have budgets for it, as you get to smaller and medium companies, AI pilots for IT can get lost in the shuffle.